Urban75 Home About Offline BrixtonBuzz Contact

Coronavirus in the UK - news, lockdown and discussion

The Guardian have taken a simplified stab at questions as to how bad this wave could be:


Towards the end we hear stuff about data in the coming period will be crucial, except the Guardian says the next week but the person they are quoting says weeks not week. And this weeks data is problematic for me because of the potential impact of the half term break last week etc. If the data is full of bad news this week then it will be useful in its own right, but if it appears to show better news then I'll be nervous about whether that actually continues or is just a temporary blip.
 
The Guardian have taken a simplified stab at questions as to how bad this wave could be:


Towards the end we hear stuff about data in the coming period will be crucial, except the Guardian says the next week but the person they are quoting says weeks not week. And this weeks data is problematic for me because of the potential impact of the half term break last week etc. If the data is full of bad news this week then it will be useful in its own right, but if it appears to show better news then I'll be nervous about whether that actually continues or is just a temporary blip.
I don't understand that model, tbh, not with the vast majority of those at high risk of hospitalisation now double-jabbed. I see the early evidence suggesting strongly the reverse. Places like Bolton and Bedford that were hit early appear, tentatively, to be past their local peaks, and without a massive surge in hospitalisation or deaths.

My back of an envelope calculation puts the effective R number reducing by around 5 per cent per week due to ongoing vaccinations, that percentage going up a bit each week as we edge towards completion. (Working on 3.5 million vaccinations a week, around 70 million more vaccinations to do the entire population, assuming that each vaccination equals 0.5 protection.)
 
Last edited:
I will try to answer you more fully a little later.

For now I'll just make a point I was going to make now anyway:

Those that were cautiously optimistic about the size of the next wave, or as some call it 'the exit wave', are aware that the estimated level of transmission advantage that the Delta variant has is one of the key variables. Their modelling numbers turn out better if it is 40% more transmissive than the Alpha variant, as opposed to say 50% or 60% more transmissible. So they were hopeful that when Hancock spoke yesterday of it being 40% more transmissible, that this was a firming up of the PHE etc estimate compared to the broader range others had spoken of up till that point.

And so I note with interest that the language used by Hancock in parliament some minutes ago was that the Delta variant appears to be at least 40% more transmissible. And so his 40% number represents the lower end of a range, not a definitive settling on 40%.
 
I don't understand that model, tbh, not with the vast majority of those at high risk of hospitalisation now double-jabbed. I see the early evidence suggesting strongly the reverse. Places like Bolton and Bedford that were hit early appear, tentatively, to be past their local peaks, and without a massive surge in hospitalisation or deaths.

My back of an envelope calculation puts the effective R number reducing by around 5 per cent per week due to ongoing vaccinations, that percentage going up a bit each week as we edge towards completion. (Working on 3.5 million vaccinations a week, around 70 million more vaccinations to do the entire population, assuming that each vaccination equals 0.5 protection.)
I'll try to be briefer than I normally am when discussing models etc with you, I'll just whack out a partial skeleton and you can choose which bits you want to discuss further.

Its not one model, its lots of models including amateur modelling attempts online. But every one of these models is capable of generating a number of very different results, and sometimes the changes to size of wave can be large with only a small adjustment to input parameters.

Possible reasons for not understanding how the hospitalisation rates can still end up so large in the current circumstances:

Not appreciating quite how large the number of daily cases could become in this wave.

Not appreciating quite how many people in ages younger than the eldest groups are still at risk of hospitalisation. A new set of data for hospitalisations in England by age group is due out in 3 days and I will no doubt use that to make some points as soon as I can.

Not appreciating how many vaccinated people are still expected to be hospitalised in the event of a significant wave of infections.

Not having factored in all of the estimates about Deltas capabilities compared to previous variants.

I dont do my own R calculations but when I get a chance I will try to point out some stuff I've seen other people doing on this front on twitter.

It is true that data from places like Bolton has given some people cause for optimism. Its too soon for me to join that club, but I have noted what they are on about. If I knew that the turnaround seen in such places will be fully sustained, and that no other location would show a worse pattern, I could join that club. Sadly it takes much longer for those matters to be settled than we've had so far, at least as far as Im concerned.

Finally I'll just say that there is significantly more uncertainty in my own view of the future and the magnitude of this new wave than I have experienced at any previous point in the pandemic. This is due to the combination of immunity through prior infection and immunity via vaccination, coupled with all the variables relating to the Delta strain. I do not exclude the possibility that this wave will be modest, but for all sorts of reasons I cannot possibly exclude the much worse possibilities either, and when I try to drill down into the detail and understant the models etc, these bad possibilities still make sense to me, they dont seem implausible.
 
And I really think you might find this blokes updated modelling thread from twitter today, that I mentioned some posts ago, to be helpful. Because he is showing a number of different scenarios based on a number of different input parameters. And goes into some detail, so I think its quite likely to offer you plenty of clues as to what factors can make models come out with either modest or large results for the coming period, or indeed some more complicated shapes and timings.

 
We're now very close to the whole of cohorts 1-9 being double-vaxed. So the vast majority of everyone over 50, health and care workers and everyone with serious preconditions. And each week, that number goes up - we'll have most people over 40 double-done by the end of the month. Pair that up with the early evidence from places like Bolton, Bedford, Glasgow, etc, which are now a month into their Delta waves and may be past their peaks, and the numbers being admitted to hospital who have had both jabs. Plus the strong evidence in all those places that infections among the young are not spreading widely to the vaccinated old. The repetition of the same pattern on these points is striking and allows some conclusions to be drawn. Add to that the fact that anywhere not currently experiencing significant rises is becoming more protected against it by around 5 per cent per week.

A model showing higher hospital admissions than we saw in January doesn't look at all credible to me given all of the above. Are they using Bayesian techniques to update their predictions as new evidence comes in?
 
My back of an envelope calculation puts the effective R number reducing by around 5 per cent per week due to ongoing vaccinations, that percentage going up a bit each week as we edge towards completion. (Working on 3.5 million vaccinations a week, around 70 million more vaccinations to do the entire population, assuming that each vaccination equals 0.5 protection.)
Leaving aside for a moment the calculation about how the vaccination programme will reduce R over time, I'd like to quickly focus on the starting point.

Have you applied that using a particular starting point for R at the moment? eg 1.4? I see that there is debate online about whether all that has changed recently has already been 'priced in' to the current estimate for R, and the only way is down, or whether some stuff hasnt fully made its presence felt yet. eg whether some of the advantages of the Delta variant have fully shown up yet or whether there is still a bit more of that to come as it completes its spread across the country. Likewise whether all the behavioural changes including those from step 3 of unlocking have fully shown up yet. And obviously a similar thing will apply if/when step 4 happens.

I can also add to that some uncertainty about seasonal effects. Some but not all modelling includes an estimate of seasonal impact on R etc, but I dont know how close to reality that stuff is, so this is yet another unclear input parameter!
 
We're now very close to the whole of cohorts 1-9 being double-vaxed. So the vast majority of everyone over 50, health and care workers and everyone with serious preconditions. And each week, that number goes up - we'll have most people over 40 double-done by the end of the month.
I dont find terms like very close, vast majority or most people to be very helpful when used in this context.

There are may different ways I can pick at this characterisation. I can point to the fact I'm 46 and I only dragged my heels booking an appointment for a few weeks, yet I'm not currently due to receive a 2nd dose until late July or early August (depending on whether they start bring forwards 2nd appointments for under 50s forwards like they have done with over 50s).

I could pick on large differences between different places, and different ethnicities. Have you looked at vaccination figures for different places in London for example? Some of it is horrific and we really need to hope that some of that is down to using wildly incorrect population figures when coming up with percentages!

And then there is stuff like this:

 
Cases now up by 52.9% over the previous week (5,683 today).

Testing numbers, which had not been updated for a few days (and were stuck on -23% odd) have been updated today, which has brought it right up again (-4.4%) following a massive increase yesterday, however over a million of those were LFT's and that's in England alone. :hmm:

No update to any of the healthcare figures, from any nation, since 4th June (last complete figures are 1/06 for patients admitted and 3/06 for patients in hospital and patients on mechanical ventilation).
 
Last edited:
Dashboard data for the London regions vaccinations is important to look at.

I shouldnt really try to post images of all the important ways they've presented this data, but I will post one indicator of the state of play in the London region for all adults. But please visit the dashboard take note of the heatmap at the bottom of the page, and the fact that there is button to choose between whether its showing first or second doses.

Screenshot 2021-06-07 at 17.01.jpg

 
Yes, of course June. :facepalm:

It's actually 12 missing from that list, I am fairly sure those were the figures Handjob announced, because I skipped back on the live coverage to double check I had typed them right. :hmm:

You're quite right - 12!
Maybe those are hospitalised children, who don't count. :thumbs:
 
A model showing higher hospital admissions than we saw in January doesn't look at all credible to me given all of the above. Are they using Bayesian techniques to update their predictions as new evidence comes in?

The models I and the media have tended to mention were presented by a number of universities to SAGE in the first days of May. I havent seen updated versions of those modelling reuslts since then. This is one of the reasons I have started to look at individual attempts to do updated modelling via twitter etc recently.

Certainly there are also people on twitter who are estimating things like R and the doubling time using the latest data. One who I've been following was hoping to see an improvement to this picture by now, and I think they are starting to get nervous, their optimism might not hold at this rate, but the week is yet young,

 
I think I understand the stuff about the effectiveness of the AZ vaccine (being too early to tell, in comparison with the earlier roll out of Pfizer?) but does any of the modelling incorporate that, too - the 60% for Delta - or does it assume complete protection after two doses of any vaccine?
 
No update to any of the healthcare figures, from any nation, since 4th June (last complete figures are 1/06 for patients admitted and 3/06 for patients in hospital and patients on mechanical ventilation).
Since the NHS alert level was reduced some time ago, they stopped publishing hospital data for England over weekends and bank holidays.

However it is still the case that to avoid additional lag in seeing hospital numbers, it is often necessary to zoom in on the figures for England and its regions only. And sometimes the data available via the NHS daily spreadsheet at the following website is more up to date than the dashboard. I believe that may be the case today but I havent actually had a chance to consume this data today yet.

In any case it is still a bit early for me to expect to see notable rises in these figures. I know I did at least one graph last week but until/unless the situation changes, I wont be doing that at high frequency.

 
So far this attempt is a failure because the ring-winger I selected is too much of a shithead to have come out with anything useful yet in the wake of the updated Delta estimates last week.

So for today at least I am taking a different approach. I've found someone else who does modelling, and I am not for the moment going to analyse where they stand on the political spectrum or their track record in this pandemic so far. Instead I am just going to drw attention to the fact they've updated their modelling with new parameters and they have posted a lengthy thread on twitter that goes into the detail. I do not have the ability to judge the accuracy of their modelling and even if their model was spot on, there is still much uncertainty about various aspects of the Delta variant which they have to feed into the model. But given that we dont currently have any more recent official SPI-M modelling than the stuff done by three academic establishments in early May, this stuff will have to do:


This is the first time I've met the term "exit wave".
 
I think I understand the stuff about the effectiveness of the AZ vaccine (being too early to tell, in comparison with the earlier roll out of Pfizer?) but does any of the modelling incorporate that, too - the 60% for Delta - or does it assume complete protection after two doses of any vaccine?
The May modelling was done before specifics of Delta were estimated. Instead the various universities picked different general scenarios to model, eg including sceenarios if a variant with x% more transmissibility came about, or if a particular variant caused a certain level of vaccine escape.

The models did use general estimates for vaccine effectiveness that existed at the time.

This is another reasons I am resorting to amateur modelling exercises, although I hope we get updated SAGE modelling made public sometime soon, perhaps in conjunction with next weeks decision.

Also modelling need to factor in the early, tentative estimates of increased hospitalisation risk from the Delta variant. (eg 2-2.5 times greater risk). I'm running out of time for right now, but check that James Ward blokes twitter modelling thread to see the detail of what he updated in his various modelling scenarios.
 
This is the first time I've met the term "exit wave".
I only heard it when I went to twitter in the last few days, and it was tending to be used by right-wingers, or optimists, or at least those who seem confident that this will be the final wave. So not a term I am hugely keen on at this stage, but perhaps it will turn out to be appropriate. I wont know for ages.
 
I ran out of time for the next few hours at least, probably just as well. But I'll just say that although the hospital data isnt highly significant yet, the regional reality does make me groan when I compare it to Hancocks recent language about hospitalisations being broadly flat (or words to that effect, I forget and dont have time to check now).
 
We're now very close to the whole of cohorts 1-9 being double-vaxed. So the vast majority of everyone over 50, health and care workers and everyone with serious preconditions.

Not sure what you mean by "very close" - I'm 56 and although I booked my first vaccination at the earliest opportunity, my second isn't until next week.

I also work with two people in their 60s who have had their first but are still waiting to be contacted with dates for the second. They both have conditions which meant they were shielding until Easter, yet are still not yet fully vaccinated.

So I'd be very cautious about making assertions like yours, unless you provide figures from a reliable source.
 
Might the lack of second vaxxes in older folks have something to do with their GPs? Because when I learned that my age range could get the jab, I signed up via the NHS website, without knowing my NHS number. I was sent to a vaccination centre in Heathrow. I was able to book a date for the second dose.

It seems weird that I would be able to get a date for my second jab when there are still people over 60 who haven't got such an appointment yet.
 
I only heard it when I went to twitter in the last few days, and it was tending to be used by right-wingers, or optimists, or at least those who seem confident that this will be the final wave. So not a term I am hugely keen on at this stage, but perhaps it will turn out to be appropriate. I wont know for ages.
Yes, it's a term that provokes some raised eyebrows from me.
 
I only heard it when I went to twitter in the last few days, and it was tending to be used by right-wingers, or optimists, or at least those who seem confident that this will be the final wave. So not a term I am hugely keen on at this stage, but perhaps it will turn out to be appropriate. I wont know for ages.
It's a bit like 'late capitalism'. Would be nice but a bit too optimistic,
 
Back
Top Bottom